Malignant gliomas are the most common primary intrinsic brain tumors and are highly lethal. The widespread migration and invasion of neoplastic cells from the initial site of tumor formation into the surrounding brain render these lesions refractory to definitive surgical treatment. Stathmin, a microtubule-destabilizing protein that mediates cell cycle progression, can also regulate directed cell movement. Nitrosoureas, traditionally viewed as DNA alkylating agents, can also covalently modify proteins such as stathmin. We therefore sought to establish a role for stathmin in malignant glioma cell motility, migration, and invasion and determine the effects of nitrosoureas on these cell movement–related processes. Scratch wound–healing recovery, Boyden chamber migration, Matrigel invasion, and organotypic slice invasion assays were performed before and after the down-regulation of cellular stathmin levels and in the absence and presence of sublethal nitrosourea ([1-(2-chloroethyl)-3-cyclohexyl-l-nitrosourea]; CCNU) concentrations. We show that decreases in stathmin expression lead to significant decreases in malignant glioma cell motility, migration, and invasion. CCNU, at a concentration of 10 μmol/L, causes similar significant decreases, even in the absence of any effects on cell viability. The direct inhibition of stathmin by CCNU is likely a contributing factor. These findings suggest that the inhibition of stathmin expression and function may be useful in limiting the spread of malignant gliomas within the brain, and that nitrosoureas may have therapeutic benefits in addition to their antiproliferative effects. [Cancer Res 2008;68(13):5267–72]
Although the initial treatment for many malignant gliomas is maximal surgical resection, the widespread migration and invasion of tumor cells into the surrounding brain tissue render a complete removal virtually impossible. Patients are therefore commonly treated with radiotherapy and chemotherapy in efforts to inhibit tumor cell proliferation and increase survival (1). The inhibition of cell migration and invasion is, in principle, likely to prevent the spread of tumors into eloquent brain regions and, in turn, preserve neurologic function. Clinical attempts to specifically effect such inhibition have, however, been limited (2).
Stathmin regulates dynamic instability, the growth and shrinkage of microtubules, by stimulating microtubule plus end catastrophes and sequestering α-β tubulin dimers (3). Consequently, stathmin has direct effects on cellular processes such as migration, division, and growth cone guidance by influencing the association of microtubules with the actin cytoskeleton (4). Consistent with this, stathmin has been shown to influence sarcoma cell migration and invasion (5). Nitrosoureas such as ([1-(2-chloroethyl)-3-cyclohexyl-l-nitrosourea]; CCNU) have previously been shown to decompose under physiologic conditions to chloroethyl carbonium ions and reactive organic isocyanates. The former alkylate DNA and cause interstrand cross-links, whereas the latter carbamoylate proteins on NH2-terminal amino groups and the amino groups of lysine side chains (6). Stathmin, which has a high percentage of lysine residues (16%), readily undergoes CCNU-mediated carbamoylation with a subsequent inhibition in microtubule depolymerization activity (7). We therefore hypothesized that CCNU, via its inhibitory effect on stathmin, could have an inhibitory effect on tumor cell migration and invasion that is independent of its antiproliferative activity.
To test this hypothesis, we first sought to establish a correlation between stathmin expression level and cell motility, migration, and invasion. We then analyzed the effects of sublethal concentrations of CCNU as well as temozolomide on these cell movement–related processes. Temozolomide was examined because of its increasing use in the treatment of patients with malignant gliomas (8). Finally, to establish a direct association between CCNU and stathmin, we assessed the effects of CCNU on the ability of stathmin to prevent the polymerization of microtubules, a well-characterized function of stathmin. The direct significance of our findings on malignant glioma therapies is discussed.
Materials and Methods
Cell culture. The U251-STMNi and U251-LacZi cell lines were derived as previously described (7). Briefly, parental U251 cells (American Type Culture Collection) were transfected with pERV3 (Stratagene) using the Lipofectamine 2000 transfection reagent (Invitrogen). Stable transfectants, selected on the basis of G418 (Invitrogen) resistance, were then transfected with pEind-RNAi, which contains the hygromycin B resistance gene and either the stathmin-specific short-hairpin RNA (shRNA) oligonucleotide sequence 5′-AGTTGTTGTTCTCTTCTATTGCCTTCTGATTGGTCAGAAGGCAATAGAAGAGAACAACAACT-3′ (U251-STMNi cells) or the LacZ-specific shRNA oligonucleotide sequence 5′-CTACACAAATCAGCGATTTCGAAAA ATCGCTGATTTGTGTAG-3′ (U251-LacZi cells), downstream of a modified muristerone A (murA)-responsive U6 promoter sequence. Cells resistant to both G418 and hygromycin B (Invitrogen) were subcloned by limiting dilution.
Cell viability assays. Cells (5 × 104) were seeded into 24-well plates (Corning, Inc.) and treated with DMSO control, 10 to 30 μmol/L CCNU (HNZ Portlink), or 10 to 30 μmol/L temozolomide (Drug Synthesis and Chemistry Branch, National Cancer Institute). After 3 d, cells were trypsinized and counted with a Coulter Particle Counter (Coulter Electronics). Experiments were performed in sextuplicate.
Immunoblotting analysis. Cells cultured for 72 h in the absence or presence of increasing concentrations of murA (Invitrogen), which enhances the activity of a modified U6 promoter and results in the transcription of downstream shRNA oligonucleotide sequences, were solubilized using 1% Nonidet P-40 in TBS (Sigma). Whole cell extracts were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore), which were subsequently probed with primary rabbit antistathmin (Calbiochem; 1:10,000 dilution) and mouse anti–β-actin (Sigma; 1:40,000 dilution) antibodies followed by secondary alkaline phosphatase–conjugated goat anti-rabbit and goat anti-mouse IgG antibodies (Promega; 1:7500 dilution). Membranes were developed using DDAO phosphate (Invitrogen) and scanned using an FLA-5100 laser based scanner (Fujifilm). Image processing including laser densitometric analysis of stathmin and β-actin expression levels was performed using MultiGauge version 3.0 imaging software (Fujifilm). Quantitative relative stathmin expression levels were determined by measuring the quantitative light absorbance units minus background (QL−BG) for each of the stathmin and β-actin bands. The ratio of stathmin(QL−BG) to β-actin(QL−BG) was determined for each sample and normalized to the stathmin(QL−BG) to β-actin(QL−BG) ratio for the nonmurA-treated sample, which was assigned the value of 100.
Scratch wound–healing recovery assays. Cells cultured in the absence or presence of 10 μmol/L murA for 72 h were allowed to reach confluence. A 20-μL pipette tip (ThermoFisher Scientific) was used to scratch and create a wound in the confluent monolayer. Detached cells were immediately removed by replacement of medium and where indicated, CCNU (10 μmol/L), temozolomide (10 μmol/L), or both were added. Images were subsequently captured, and then again 24 h later, using an AxioCam MRm camera, an Axiovert 200 microscope, and AxioVision version 4.2 software (all Zeiss). Experiments were performed in triplicate.
Cell migration assays. Cells cultured in the absence or presence of 10 μmol/L murA for 72 h were added to the upper compartments of 96-well Boyden chamber plates containing 8-μm pore size polycarbonate filters (Neuro Probe). After 16 h of culture in medium containing drug vehicle, CCNU (10 μmol/L), temozolomide (10 μmol/L), or both CCNU and temozolomide, the polycarbonate filters were removed, fixed with methanol for 10 min, and stained with modified Trypan Blue stain (0.2% in PBS) for 20 min to facilitate visualization and counting of cells. Cells on the upper side of the filter were removed using a rubber scraper, and the migrated cells on the underside of the filter were counted with the aid of a microscope. Experiments were performed in sextuplicate.
Matrigel invasion assay. The polycarbonate filter of each Boyden chamber well was coated with Matrigel (BD Biosciences). Cells cultured in the absence or presence of 10 μmol/L murA for 72 h were inoculated into the upper chambers, and the chemoattractant rhSDF-1α/PBSF (pre–B-cell growth–stimulating factor, 100ng/mL; R&D Systems) was added to the bottom chambers. After 24 h of culture in medium-containing drug vehicle, CCNU (10 μmol/L), temozolomide (10 μmol/L), or both CCNU and temozolomide, the Matrigel-coated polycarbonate filters were removed, fixed, and stained as above. Invaded cells on the underside of the filter were counted as above. Experiments were performed in sextuplicate.
Glioma invasion in organotypic brain slices. Brain slices were obtained from 10-d-old NIH Swiss mice and cultured as previously described (9). Glioma cells incubated with DiI (3H-Indolium, 2-[3-(1,3-dihydro-3,3-dimethyl-1-octadecyl-2H-indol-2-ylidene)-1-propenyl]-3,3-dimethyl-1-octadecyl-, perchlorate; 30 μmg/mL; Invitrogen) were plated onto 6-well plates at a density of 5 × 104cells/10 μL. Upside-down overnight culture of the cell suspensions led to the formation of glioma spheroids ∼0.5 mm in diameter. Spheroids were picked and implanted into the corpus callosa of the brain slices with the aid of an inverted microscope equipped with a calibrated micromanipulator. Slices were maintained in culture medium containing no drugs, CCNU (10 μmol/L), murA (10 μmol/L), or both CCNU and murA. Medium, including drug where appropriate, were replenished every 3 d. Images were captured after 4 wk and analyzed using the same equipment and software as for the scratch wound–healing assays. Animal studies were carried out in accordance with NIH institutional guidelines for the care and use of animals. Experiments were performed in triplicate.
In vitro tubulin polymerization assay. Tubulin polymerization assays were performed using purified rat tubulin as described (10, 11). Six-histidine–tagged recombinant stathmin (5 μmol/L) was produced as previously described (7), and when indicated, was pretreated for 6 h with CCNU (10 μmol/L) followed by removal of free CCNU using a Zeba micro desalt column (Pierce). Experiments were performed in triplicate.
Statistical analysis. Student's t tests were performed where appropriate using JMP 5.1 (SAS Institute, Inc.). P value of <0.05 was considered significant.
Stathmin knockdown and CCNU inhibit scratch wound healing. To determine the effects of stathmin knockdown on cell motility, we first generated a U251 malignant glioma–derived cell line in which murA enhances the transcription of a stathmin-specific shRNA and induces the knockdown of stathmin expression (U251-STMNi). We also generated a control cell line in which murA enhances the transcription of a LacZ-specific shRNA (U251-LacZi). The murA-induced knockdown of stathmin protein expression by U251-STMNi cells is dose dependent with a 47% decrease occurring in response to treatment with 10 μmol/L for 72 hours (Fig. 1A). In contrast, treatment of U251-LacZi cells with murA concentrations as high as 20 μmol/L do not have appreciable effects on stathmin expression (Fig. 1B). All further experiments involving the use of murA were carried out using a concentration of 10 μmol/L.
Given the cytotoxic effects of CCNU and temozolomide, we sought to establish experimental conditions under which decreased cell motility rather than decreased cell viability would be the primary potential cause for decreased scratch wound healing. Treatment of cells with 10 μmol/L CCNU does not have a significant effect on cell viability (P = 0.6520); however, treatment with 20 μmol/L CCNU leads to a 28% decrease in cell number after 72 hours (P = 0.0100; Fig. 1C). Similarly, treatment of cells with 10 μmol/L temozolomide does not have a significant effect on cell viability (P = 0.5935); however, treatment with 20 μmol/L temozolomide leads to a 17% decrease in cell number after 72 hours (P = 0.0411; Fig. 1C). All further experiments involving CCNU and/or temozolomide were therefore carried out using a concentration of 10 μmol/L to minimize the potential confounding effect of decreased cell viability.
U251-STMNi cells treated with murA alone (Fig. 2A,3) show decreased wound healing compared with nonmurA-treated U251-STMNi cells (Fig. 2A,2). As controls for the effects of murA treatment and shRNA expression per se, U251-LacZi cells with (Fig. 2B,3) and without (Fig. 2B,2) murA pretreatment were also assessed, and there seems to be no differences in wound healing. CCNU treatment alone inhibits wound healing by both U251-STMNi (Fig. 2A,4) and U251-LacZi (Fig. 2B,4) cells, but temozolomide treatment alone does not affect either cell type (Fig. 2A,5 and B5). The combination of stathmin knockdown and CCNU treatment seems to be additive in inhibiting U251-STMNi (Fig. 2A,6) but not U251-LacZi (Fig. 2B,6) cell wound healing. There is no further decrease in wound healing in either cell type due to the addition of temozolomide to murA or CCNU (Fig. 2A 78, A8, B7, and B8).
Stathmin knockdown and CCNU inhibit cell migration and invasion. To quantitate the effects seen in the scratch wound–healing assays, Boyden chamber migration assays were performed (Fig. 3). MurA induced knockdown of stathmin expression in U251-STMNi cells causes a 32% decrease in migration (P = 0.0008), whereas CCNU treatment causes a 35% decrease (P < 0.0001). Treatment of U251-STMNi cells with the combination of murA and CCNU causes a 62% decrease in migration, which is significantly greater than that seen after treatment with the former (P = 0.0002) or the latter (P = 0.0004) alone. In contrast, temozolomide treatment by itself has no significant effect on migration (P = 0.1092) and does not enhance the effects of murA (P = 0.6701) or CCNU (P = 0.4027) either (Fig. 3A). CCNU treatment of U251-LacZi cells causes a 30% decrease in migration (P = 0.0008), but murA treatment of these cells has no significant effect when used alone (P = 0.5960) and does not enhance the effects of CCNU (P = 0.1297). Temozolomide alone also has no significant effects on the migration of U251-LacZi cells (P = 0.3839) and also does not enhance the effects of CCNU either (P = 0.1138; Fig. 3B).
The movement of cells through Matrigel, an in vitro model of basement membrane invasion, was evaluated. In U251-STMNi cells, murA-induced stathmin knockdown inhibits invasion by 49% (P < 0.0001), whereas CCNU treatment does so by 50% (P < 0.0001). The combination of the two inhibits invasion by 70% (P < 0.0001), which is significantly greater than that seen in response to either stathmin knockdown (P = 0.0013) or CCNU treatment (P = 0.0014) alone (Fig. 4A). CCNU treatment of U251-LacZi cells causes a 40% decrease in invasion (P = 0.0003), but murA treatment of these cells has no significant effects (P = 0.0756; Fig. 4B). Again, temozolomide has no effects by itself and does not enhance the effects of either murA treatment or CCNU treatment in either U251-STMNi or U251-LacZi cells (all P > 0.0821; Fig. 4A and B).
We also assessed the movement of tumor cells within organotypic slice preparations, a model for the in vivo invasion of tumor cells into surrounding brain tissue. As expected, U251-STMNi cell invasion is decreased by murA, CCNU, and the combination of murA and CCNU (Fig. 5A). U251-LacZi cell invasion is also decreased by CCNU, but murA treatment does not by itself seem to have any appreciable effects nor enhance the effects of CCNU in this cell type (Fig. 5B).
CCNU directly inhibits stathmin function. Stathmin prevents microtubule polymerization by sequestering α-β tubulin dimers (3). To determine if CCNU could inhibit this stathmin function, tubulin polymerization assays were performed. The steady-state concentration of polymerized microtubules in a tubulin-containing solution is decreased 63% by the addition of native stathmin. This stathmin-mediated decrease is abrogated by 59% if the stathmin is pretreated with 10 μmol/L CCNU (Fig. 6). CCNU by itself has no significant effects on tubulin polymerization (data not shown).
The infiltration of malignant gliomas into eloquent brain regions compromises neurologic function and is a major cause of morbidity. Although most malignant glioma therapies are intended to be cytostatic or cytotoxic, therapies intended to inhibit tumor cell migration and invasion are also likely to be beneficial. With this in mind, we investigated the roles and interactions of stathmin and CCNU in the directed movements of malignant glioma cells. The results presented above show that both decreases in stathmin expression and treatment with CCNU can inhibit the migration and invasion of malignant glioma cells. The concentration of CCNU used throughout these studies (10 μmol/L) was specifically selected because it has no significant effects on cell survival and thereby minimizes decreased cell viability per se as a potential confounding effect on cell motility related functions. Ten μmol/L is also the approximate mean peak plasma concentration achieved in patients after p.o. administration of CCNU at a dose of 15 mg/kg (12).
A plausible mechanism by which CCNU can retard migration and invasion is through the direct inhibition of stathmin function as shown in the tubulin polymerization assay. Consistent with this model, temozolomide, an imidazotetrazine DNA alkylating agent with no direct effects on stathmin function (7), has no effects on cell migration or invasion. Temozolomide also does not act to enhance the inhibitory effects of stathmin knockdown or CCNU treatment on these cell movement related functions.
The median survival of malignant glioma patients treated with radiotherapy plus temozolomide has been shown to be significantly greater than that of patients treated with radiotherapy alone (P < 0.001; ref. 13). In contrast, the addition of nitrosoureas to radiotherapy has not been shown to provide a significant overall survival benefit (P = 0.108; ref. 1). One implication of these two studies is that temozolomide is more effective than nitrosoureas in inducing lethal DNA cross links in proliferating tumor cells. Although this is likely the case, our results indicate that nitrosoureas are superior to temozolomide in inhibiting tumor cell motility. A possible therapeutic strategy would then be radiotherapy plus the combination of temozolomide and a nitrosourea, the former to control proliferation and the latter to control invasion.
In a recent phase II clinical trial of radiotherapy and temozolomide plus CCNU, malignant glioma patients whose tumors had promoter methylation of the O6-methylguanine-DNA methyltransferase (MGMT) gene had a median progression-free survival of 19 months (14). In a phase III clinical trial of patients also with MGMT promoter methylated tumors, the median progression-free survival after radiotherapy and temozolomide treatment was 10.3 months (13). The almost 2-fold increase in progression-free survival with the addition of CCNU is consistent with a model in which progression is due to a variable combination of local and/or distant tumor recurrence defined as expanding and/or new areas, respectively, of contrast enhancement on magnetic resonance imaging. The latter requires the movement of tumor cells away from the original tumor site. Due to its inhibitory effects on migration and invasion, CCNU may delay this infiltration of tumor cells into surrounding brain tissues and, consequently, the appearance of new areas of contrast enhancement. When eloquent brain regions are spared, there may also be a delay in the deterioration of neurologic function. A phase III clinical trial assessing the addition of nitrosoureas to radiotherapy and temozolomide should therefore be considered, given our experimental data and the promising results of the phase II clinical trial described above.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Note: X-J. Liang and Y. Choi contributed equally to this work.
Grant support: Intramural Research Program of the NIH, National Institute of Neurological Disorders and Stroke and National Institute of Child Health and Human Development.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank A. Sedlock for providing technical assistance, G. Park for critical review of the manuscript, and L. Wu and J. Isaac for advice on the organotypic slice cultures.